2026-05-06T09:27:21.210Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:7779 2020-12-13T14:45:39Z 1

Anomalies Detection in the Application Logs Using Kohonen SOM Machine Learning Algorithm 2020 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/7779 http://portal.sinteza.singidunum.ac.rs/Media/files/2020/275-282.pdf V. Marković A. NJeguš М. Марјановић-Јаковљевић Internal fraud in the financial sector is difficult to detect since fraudulent transactions are indistinguishable from ordinary transactions, and standard checkpoints, in the form of transaction documentation and authorization, are skillfully avoided. Well-designed software has available and machine-readable application logs that can be analyzed to detect anomalies in application usage. This paper presents a data preparation technique using path analysis and Kohonen SOM clustering algorithm that can help better profile users of an application to reduce the number of cases that will be further investigated. conferenceObject 275 282 doi:10.15308/Sinteza-2020-275-282 Proc. of International Scientific Conference on Information Technology and Data related research